SAN JOSE, Calif. – Advanced Micro Devices Inc. has unveiled a new lineup of accelerator chips in an effort to compete with Nvidia Corp. in the artificial intelligence market. The company announced the MI300 at an event in San Jose, California, where CEO Lisa Su predicted that the AI chip industry could reach over $400 billion in the next four years, more than double AMD’s previous forecast.

The MI300 launch represents a significant move for AMD in the AI accelerator market, setting the stage for a showdown with Nvidia. These accelerator chips are crucial in developing AI models, as they handle data bombardment more efficiently than traditional computer processors. AMD is targeting big names in technology such as Microsoft Corp., Oracle Corp., and Meta Platforms Inc. as customers for the MI300 lineup.

Nvidia’s stock dipped in response to AMD’s announcement, highlighting the growing competition in the AI accelerator market. However, Nvidia remains a dominant force, with significant demand for its chips from data center operators. Despite this, AMD sees an opportunity to carve out a share of the market, particularly in the area of large language models, where it believes its advantage lies in computer memory capacity.

The new AMD chip boasts over 150 billion transistors and significantly more memory and memory bandwidth than Nvidia’s current market leader, the H100. AMD claims that its new chip is on par with Nvidia’s H100 for training AI software and superior in terms of inference – the process of running the software for real-world use. However, Nvidia is also developing its next-generation chips, with the H200 set to succeed the H100 and provide access to new high-speed memory.

AMD’s bullish prediction of a $400 billion AI processor market reflects the optimism in the industry, which is growing rapidly. However, it will take time for AMD to capture a significant share of this market, with estimates indicating that the chipmaker’s revenue from accelerators will surpass $2 billion in 2024.

The chips are based on graphics processing units (GPUs), traditionally used by video gamers. Their ability to rapidly perform calculations simultaneously has made them the preferred choice for training AI software.

In conclusion, AMD’s unveiling of the MI300 lineup represents a significant development in the AI accelerator market, signaling increased competition with Nvidia. The competition between these two tech giants will likely drive innovation and advancements in AI hardware, benefiting the industry as a whole. Despite Nvidia’s current dominance, AMD’s ambitious forecast demonstrates its determination to carve out a significant share of this rapidly growing market.